Research On Storage And Recognition Technology Of Wheat Disease Image

Wheat is one of the most important food crops in China, and it is easy to be affected by environmental factors in the process of wheat growth, which affects the yield of wheat. The wheat disease image recognition and storage efficiency low, this paper presents the local Support Vector Machine(SVM) classification algorithm for wheat disease image identification, and the improvement of the local Support Vector Machine algorithm, proposes the use of particle swarm algorithm and technical data of local Support Vector Machine is improved, classification accuracy and the efficiency of optimization. According to the storage of wheat disease image, this paper puts forward the technology of using big data to store the image of wheat diseases, and improve the efficiency of storage.Support Vector Machine is based on statistical learning theory based on kernel machine learning method, it in category prediction of each sample, the use of is training set all the sample information(global information) cannot make full use of the local information in the sample. Local Support Vector Machines can make full use of local information in the sample. But in the process of classification of local Support Vector Machine as well as its deficiency and local Support Vector Machine in the process of classification is not able to fully utilizes the attributes of sample points, in the classification will be the weight of each attribute set is the same, so cannot reflect the importance of different attributes in classification. In this paper, a local Support Vector Machine algorithm based on particle swarm optimization is proposed, which is used to optimize the function of each attribute in the classification.Big data is a new research field in recent years, Hadoop parallel computing platform is a parallel computing technology in the big data environment. Hadoop parallel computing can reduce the time complexity of the algorithm, the algorithm is changed to parallel computing by the process of serial computing. The improved local Support Vector Machine with Hadoop parallel computing technology combined to solve based on particle swarm of local Support Vector Machine classification time complex degree high problem. Experiments show that Hadoop platform based on particle swarm local Support Vector Machine algorithm can reduce the classification time complexity while keeping almost the same classification accuracy.HBase database is based on Hadoop platform a NoSQL database, compared to conventional SQL database HBase can provide high concurrent read / write operations support, can be stored in a non-structured data. Therefore, this paper uses the HBase database to the acquisition and storage of wheat disease image information, to improve the efficiency of image storage, and is convenient to use Hadoop platform for image data processing.In order to verify the improvement of local Support Vector Machine in the efficiency of wheat disease image recognition, respectively, using local Support Vector Machine and based on particle swarm of local Support Vector Machine algorithm in image data of wheat disease concentration test and comparison experiments. The experimental results show that, in general don't accuracy we improved the local support vector confidential is better than ordinary local Support Vector Machine.